Confidence assessment
A confidence assessment is a stated probability that a Key Result will land, given by the person closest to the work and updated every week. Its job is not accuracy for its own sake. Its job is to make a falling number visible while it is still cheap to act on.
The obvious bit
Progress tells you where you have got to. Confidence tells you where you are going to end up. Only one of those is actionable in week four.
A Key Result at forty percent of target in week six is not information yet. At forty percent with confidence falling from 80 to 40, it is a decision waiting to be made.
How we run it
One number per Key Result, weekly, from the owner. We use a simple scale rather than false precision: high, medium, low, or a percentage if the team prefers, with agreed meanings written down once so people are not silently using different definitions.
The number is a forecast of the ending, not a rating of the effort. And the only interesting thing about it is the direction of travel. A steady 60 is fine. A 90 that becomes a 50 over three weeks is the most valuable signal in the review, and it is the reason the assessment exists at all.
A drop is not a confession. It is the system working, and it should be treated as such out loud, because people learn from what they watch happen to whoever reports first.
Why it is the honest half of a check-in
Most check-ins ask about progress, which is a question about the past and safe to answer. Confidence asks about the future, which requires a judgement and cannot be satisfied with activity.
That is precisely why it is the question worth asking. A team can be busy and behind at the same time, and progress reporting hides that. Confidence does not.
The failure modes
Confidence as a grade. The moment a low score costs someone something, everyone reports high until the last possible week. You will have destroyed the one signal that gave you time.
Confidence theatre. Everything at 70 forever. Usually a sign the scale has no agreed meaning, or that nobody believes the number changes anything.
Precision instead of honesty. Arguing about 65 versus 70 is a way of avoiding the real question, which is whether this lands at all.
Confidence without a response. If a falling score does not trigger a conversation about resource, scope or stopping, people quickly learn it is paperwork.
Our opinion
Confidence assessment is the cheapest thing on this list to introduce and the fastest to break. It costs a minute a week per Key Result and it will tell you more about your culture in a quarter than any survey.
If your scores never move, you do not have a measurement problem. You have an honesty problem, and the score is telling you about it anyway.
One line to keep: a confidence score is an early-warning system, so the moment it becomes a performance measure it stops working.
Who sets the confidence score? +
The person closest to the work, not their manager. A score set upward is a negotiation; a score set by the owner is a forecast.
What should trigger action? +
A sustained fall, not a single dip. Two or three consecutive weeks of decline is a conversation about scope, resource or stopping, and it should happen in the check-in rather than at quarter end.
Should confidence affect anyone’s appraisal? +
No, and this is the one rule we do not bend. The instant confidence carries a personal cost, everyone reports high and you lose the early warning the whole practice exists to give you.
From the ZOKRI OKR Handbook, the methodology we install and maintain. Written by Matt Roberts.
A UK-based strategy and OKR consultant and two-time SaaS founder with a venture-backed exit, Matt turns strategy into execution for teams scaling from tens to thousands. He co-founded ZOKRI in 2018, having previously co-founded Linkdex, a venture-backed enterprise SaaS platform he led to a trade sale. He writes the methodology behind these notes.
Confidence assessment is one of the first things we install, because it changes what a check-in is for. We set the scale, coach the first quarter, and make the first honest drop safe.